Contributor Guidelines

Operator: EgoVista SAS, registered with the Paris Trade and Companies Register under SIREN 106 489 735, 47 rue Vivienne, 75002 Paris, France. Contact: contact@egovista.app | Data protection: privacy@egovista.app.

What this document is. These Guidelines set out what we ask of you when you record and submit a video, and the reasons a video may not be accepted. They form an integral part of the Contributor Terms of Service. Every rejection notice we send you names one of the reasons listed here and links to the section that explains it. If a reason is not in this document, it is not a reason we use.

How to read it. Sections 2 and 3 are about rights and permissions, and they matter most: a problem there is not a quality problem, it means we cannot lawfully use the footage at all. Sections 4 and 5 are about the recording itself and about the mission you accepted. Section 6 explains how decisions are made and what happens when a video is not accepted.

What we ask of you and nothing more. You choose when, where and whether to record. There is no schedule, no minimum, and no penalty for not submitting. These Guidelines describe the properties a video must have to be usable, not a method you must follow.

Video, and anything else a mission may ask for. These Guidelines are written for missions that ask for video, which is what almost every mission asks for. The Terms call what you deliver a Submission, because a mission may ask for material other than, or in addition to, video. Where this document says video, read the material you submitted under the mission concerned.

Where a mission asks for something other than video, its own page says so, states what that modality requires, and nothing is part of that mission that is not written on its page. Sections 2 and 6 of this document apply whatever the modality: the rights you must hold and the way decisions are taken do not change with the format.

1. Who can contribute

1.1 You must be 18 or over. This is an absolute condition, with no exception in any country. We ask you to confirm your age at signup, and we may ask for documentation to verify it.

1.2 One account per person. Operating more than one account is a breach of the Terms and ends the relationship without compensation. This is the one rule in this document whose consequence goes beyond a single video.

1.3 Where you must live. The list of countries from which we can accept contributions and make payments is published on the signup page and can change. Payment routes differ by country and some are not available to us. If your country is not on the list at the time you sign up, we cannot pay you, and we will say so before you record anything.

1.4 The equipment is yours. A recent smartphone, a GoPro, or smart glasses, with an inexpensive mount. We do not supply equipment and we do not require any particular brand or model. Each mission states the minimum resolution and frame rate it needs, and you check that your device can meet them before you accept it.

2. Before you record: rights and permissions

Everything in this section exists for one reason. We license the footage you send us to professional buyers. If you did not hold the rights to a video, we cannot pass those rights on, and neither we nor you can fix that after the fact. This is why a problem in this section means no payment and no retention of the file, and why we cannot make exceptions.

2.1 You must be the person who recorded the video. You film it yourself, with your own device. Footage recorded by someone else, downloaded, bought, found online, generated, or taken from any source other than your own camera cannot be submitted, even with that person's permission. We are licensing your work, and only you can grant that.

Reason code: the video was not recorded by you, or you do not hold the rights to license it.

2.2 A video is submitted once, to one buyer. You keep full ownership of your raw footage and you remain free to record for anyone else, including our competitors. What you cannot do is deliver the same recording twice, whether to us twice or to us and to another platform. Each video you submit must be a recording that has not been delivered anywhere before.

Reason code: this video has already been delivered, here or elsewhere.

2.3 Other people in your video. Anyone who appears in your video as a subject, meaning deliberately filmed rather than passing through the frame, must have signed an image release before you record. The template is in Annex 1 of the Terms, and we may ask you for a copy at any time. Someone identifiable with no signed release covering them means the video cannot be used.

There is a second situation, and it is not your fault when it happens. Sometimes a person is positioned in such a way that we cannot blur them without destroying what makes the footage useful, for example when their face occupies the working area you are filming. We will tell you this is what happened, and you are paid.

Reason codes: someone other than you is identifiable and no signed image release covers them. A person appears in a way that cannot be occluded while keeping the video usable.

2.4 Telling people what the recording is for. This is an obligation, not a courtesy, and it applies to everyone in your household or workplace who may appear in your footage, including people who only pass through the frame. Before you start recording, you tell them three things: that you are recording first-person video, that the footage will be processed and sold as part of an annotated dataset used to train artificial intelligence and robotics models, and that they can refuse. A template you can read out or send is in Annex A of this document.

If a person objects, you do not film them, and you do not film in a place where they will be in frame. There is no version of this where you record first and ask later.

Reason code: a person filmed as a subject was not told what the recording would be used for, or objected to being filmed.

2.5 Where you record. Your own home and public spaces where filming is permitted. Anywhere else belonging to someone else, a private residence, a private office, a shop, a restricted area, a workplace that is not yours, requires the written authorisation of whoever runs it, obtained before you record and available to us on request. Changing rooms, restrooms, medical facilities, schools and religious sites are never acceptable, with or without authorisation.

Your own workplace is a case of its own. Recording in a professional setting, meaning a place where you work, where customers come, or where other people are employed, is possible only under a mission that expressly provides for it and that carries Particular Conditions, which you accept when you join it. Outside such a mission, a professional setting is not a place you may record for us, even when the place is yours.

Reason codes: the video was recorded in a place you were not authorised to film. The video was recorded in a professional setting under a mission that does not provide for it.

2.6 No one under 18, ever. No person under 18 may be identifiable in your footage, as a subject or incidentally. This is not a matter of blurring or of parental permission. If a child may enter the frame, record at a different time or in a different room. If one appears in a recording you have already made, do not submit it.

Reason code: a person under 18 is identifiable in the video.

2.7 What must never be in frame. Some things cannot be collected lawfully whatever permission you have, and blurring does not solve them. Do not record documents, screens or surroundings that reveal a person's health, religion, political views, sexual orientation, or trade union membership. Do not record the inside of a private space belonging to someone else. Do not record anything unlawful, violent, sexual, hateful, or that infringes a third party's intellectual property, including screens showing copyrighted films, games or software.

Three further things belong to someone other than you and must not appear: anything belonging to a customer of yours, anything covered by an obligation of confidentiality you have given, and anything that is a trade secret of a third party. Where the trade secret is your own, you may film it, and by submitting you allow us and our buyers to use what is shown.

Reason codes: the video reveals health, religious or political information, or the inside of someone else's private space. The video contains content that cannot be lawfully collected or distributed. The video shows material belonging to a customer, covered by confidentiality, or constituting a third party's trade secret.

2.8 If someone changes their mind. A person who signed a release can withdraw it at any time, and so can a person who agreed verbally to being in frame. When that happens after you have submitted, tell us at privacy@egovista.app. We remove the footage. This is not a fault on your part, and it does not affect anything else you have submitted.

Reason code: someone who appears in the video has withdrawn their authorisation.

2.9 If you record with smart glasses. Glasses are discreet, and that is precisely the problem the French data protection authority has asked platforms to address. Three rules follow from it. Keep the recording indicator visible and never cover it. Never record in an intimate space, which includes changing rooms, restrooms and medical facilities. Never use the discreetness of the device to record someone who would have objected if they had known. If you would not have recorded the same scene with a phone held up in front of you, do not record it with glasses.

Reason code: the recording indicator was hidden, or the video was recorded in a private space such as a changing room, a restroom or a medical facility.

2.10 People who can be heard. A person whose speech is audible and understandable is in your recording as much as if they were in frame, and section 2.4 applies to them in full. Record when no one is talking nearby, or tell them first and let them refuse.

This is a question of that person's rights, not a question of your video's quality. The presence of sound in your file is never itself a problem: see section 4.6.

Reason code: a person who was not told what the recording was for is audible and understandable in it.

3. What the mission asks

Each mission has its own page, and that page states what it requires: the environment, the length, the minimum resolution and frame rate, the mission code, and the opening movement. Those requirements are fixed when the mission opens and do not change afterwards. Read the page before you record, not after.

3.1 The mission code. When you join a mission, we issue you a short code, shown on the mission page. It is yours and it belongs to that mission. Show it clearly to the camera at the start of your recording, held steady and in focus for two or three seconds, before you begin the task.

The code exists so that a recording can be tied to the mission it was made for. It is not a security check and nothing verifies it automatically. A reviewer looks at your footage and sees it, or does not.

If we failed to issue you a code, or the mission page did not display one, the absence of a code is our problem and not yours, and you are paid.

Reason code: the code issued for this mission is not clearly visible in the video.

3.2 The opening movement. Each mission asks for one specific hand movement, described on the mission page, performed before you start the task, with your hand fully in frame. It takes a few seconds. Like the code, it is checked by a person and not by software, and if the mission page did not tell you which movement to perform, its absence is our problem.

Reason code: the movement this mission asks you to perform is not in the video.

3.3 The environment. A mission asks for a specific kind of setting, a kitchen, a workshop, a garage. Recording a different setting means the footage does not answer the brief, however good it is.

Reason code: the environment in the video is not the one this mission asked for.

3.4 Length, and what counts as usable. Each mission states a minimum and a maximum, and both apply to usable footage rather than to the raw file. Usable means the part where the task is actually being performed with your hands in frame. Time spent setting up, walking to the room, or standing still does not count towards the minimum, and it is not paid.

The usable length is measured on your file, and you are told what was retained and which parts were not counted. Each mission page states how much shorter the usable length may be than your file before a person looks at the measurement rather than accepting it. If you think the measurement is wrong, section 6.6 is how you say so.

Reason code: the usable length of the video is outside the range this mission requires.

4. Recording quality

4.1 First person, your own hands. The camera is on your head or your chest, looking where you look. Your own hands do the task and are visible while doing it. A video filmed from a tripod, held at arm's length, pointed at someone else working, or where the hands are mostly out of frame, is not egocentric footage and cannot be used for what we build.

Reason code: the video is not a first-person recording of your own hands.

4.2 Resolution and frame rate. Each mission states its minimums. Check your camera settings before you record rather than after, because resolution cannot be recovered afterwards. If a mission was open to a device that could not meet its own stated minimum, that is a mistake on our side and you are paid.

Reason code: the video resolution or frame rate is below what this mission requires.

4.3 A legible image. We need to see what your hands are doing. Enough light on the working area, focus held on it, and no motion blur so heavy that the objects smear. Backlighting, a dark room, a lens smeared with fingerprints, or a camera that keeps hunting for focus all make footage unusable even when everything else is right. Wipe the lens, add light, and watch the first ten seconds back before you record the whole thing.

Reason code: the working area is not clearly visible: exposure, focus or motion blur.

4.4 Steady enough to follow. Head-mounted cameras move, and that is expected. What we cannot use is motion so fast or so constant that an object cannot be tracked from one moment to the next. Turn your head deliberately rather than sharply, and pause a beat when you change what you are looking at.

Reason code: camera motion makes the manipulated objects impossible to follow.

4.5 One continuous take, nothing added. Submit the file as your camera produced it. No cuts, no joining two takes, no speeding up or slowing down, no stabilisation applied afterwards, no filters, no text, no logos, no added music. If a take goes wrong, record it again and submit the good one. Editing breaks the continuity of the movement, which is the thing being measured.

Reason code: the video has been edited, cut, sped up, or has overlays or added music.

4.6 Sound. Missions concern the image. Cameras record sound by default and no reasonable instruction reliably prevents it, so the presence of a sound track in a file submitted under a mission that did not ask for it is never a fault on your part and is never a reason to refuse a video. Outside a mission that expressly asks for sound, we do not exploit it, do not annotate it, and do not include it in any dataset delivered to a buyer.

What does matter is who can be heard. That is a question of rights and it is dealt with in section 2.10.

Where a mission does expressly ask for the sound track, its page says so and says why, and everything in section 2 about telling people applies to anyone who is audible and not only to anyone who is visible.

5. Submitting the file

5.1 A complete upload. Large files fail on unstable connections. Stay on the page until the upload confirms, and prefer a fixed connection over mobile data for long recordings. If an upload does not finish, nothing is lost and nothing is held against you. Start it again.

If a file arrives incomplete because of a fault on our side, we treat it as ours and you are paid.

Reason code: the video file did not finish uploading, or is missing.

5.2 A readable file. MP4 or MOV, straight from the camera. Exotic formats, files exported by editing software with unusual settings, and files damaged during transfer cannot be opened, and a file we cannot open cannot be reviewed.

Reason code: the file could not be read: unsupported format, or the video stream is damaged.

5.3 What you declare at submission. Before you send a video, we ask you a small number of questions: whether anyone else appears in it or can be heard in it, whether you told them, what kind of place you recorded in, and whether anything in it belongs to a customer, is covered by confidentiality, or is someone else's trade secret. Answer them for the video in front of you, not from habit. These answers are recorded, dated, and kept as they were given.

Answer them accurately even when the accurate answer is inconvenient, because a declaration that does not match the footage costs us more than the problem it was hiding. Where a difference between what was declared and what we observe could be explained either way, we resolve it in your favour and ask you rather than rejecting.

Reason code: what was declared about this video does not match what we could observe.

6. How decisions are made

6.1 Who decides. A submission is assessed against the requirements published in this document and on the mission page, and the usable footage it contains is measured.

A refusal is never taken by software alone. Whenever the assessment points to a refusal, a person looks at your video and takes the decision.

A person also looks at it, and decides, in three further cases: when it is the first submission you have made, when the usable length retained is shorter than your file by more than the proportion published on the mission page, and when you contest a decision under section 6.6.

Outside those four cases, a submission may be accepted automatically. When that happens you are told the usable length retained and shown which parts of your file were not counted, and you can contest both under section 6.6, where a person will look at it.

6.2 The three outcomes. A submission is accepted, in which case it is paid for its usable length, whether the acceptance was taken by a person or automatically. Or it is not accepted, in which case a person has decided so and you receive the reason from the list in this document, naming the section that explains it. Or it is held for a closer look, which simply means the decision takes longer.

6.3 What a rejection is and is not. A video that is not accepted is not paid. That is the whole of the consequence. Your account stays open, your access to other missions is unchanged, nothing is recorded against your name, and you may record and submit again for the same mission. We do not rank contributors, and a rejection does not make the next decision stricter.

6.4 When it is not your fault. Some reasons in this document describe a failure on our side rather than on yours: a pipeline that did not run, a brief that did not say what it should have, a mission that asked for something its own requirements made impossible. When the reason we give you is one of those, you are paid in full for the submission. Where fault genuinely cannot be established, the benefit of the doubt goes to you.

Two situations sit outside both: a video we have no right to use, and a video that duplicates one already delivered. Those are neither paid nor kept, whatever the intention behind them.

6.5 Payment. Missions are paid per hour of usable footage, at the rate stated on the mission page before you accept it. Usable footage is defined in section 3.4. The rate is per mission and is never a wage or an hourly rate for time spent.

6.6 If you disagree with a decision. Reply to the decision notice, or write to contact@egovista.app, stating the video and why you think the decision is wrong. This covers a refusal and it covers the usable length retained.

A person reads what you write, considers the point you make, and answers within thirty days. The decision can be changed, and where it is, you are paid as if the right decision had been taken the first time. Where the decision you contest was taken automatically, this is how you obtain a human decision on it. Data protection questions go to privacy@egovista.app.

7. Changes to this document

These Guidelines are versioned. The current version is v1.1, effective 7 August 2026. The text of each version is published at a stable address so that anyone can check it has not been altered after the fact, alongside a fingerprint of its content.

Changes apply to missions opened after the date a new version takes effect, never to submissions already made. A decision taken on your video always cites the version in force when you accepted that mission, and we cannot re-examine a past submission under a later version.

These Guidelines form an integral part of the Terms and are not accepted separately from them. A version of these Guidelines binds you from the moment you accept the version of the Terms that incorporates it. Where a change is imposed without being submitted for acceptance, it is notified by email at least thirty days before it takes effect.

The present version takes effect on 7 August 2026, the same day as version 2.3 of the Terms, in which it is incorporated. No mission had been opened under version 1.0, so no submission and no decision is affected by the change.

What changed from v1.0. Section 6.1 now describes automated acceptance, which version 1.0 excluded: a refusal is still never automatic, and three further cases still go to a person. Section 4.6 no longer treats the presence of sound as a reason to refuse a video, and the question of who can be heard has moved to section 2.10, where it belongs. Section 2.5 states the condition under which a professional setting may be recorded. Section 2.7 and section 5.3 cover material belonging to a customer, covered by confidentiality, or constituting a third party's trade secret. Section 3.4 says how the usable length is measured and what you are told about it. Annex A no longer states that all processing happens inside the European Union.

Annex A — Telling people what the recording is for

Read this out, send it as a message, or say the same thing in your own words. What matters is that the three elements are covered and that the person had the chance to refuse before you started.

I am going to record video from a camera on my head. It films what I see and what my hands are doing, and you may appear in it.

The video goes to a European company called EgoVista, which processes it and sells it to laboratories and companies that use it to train artificial intelligence and robots. Faces are blurred before the video leaves their systems. The video is stored and the main processing runs in the European Union; some of the technical providers involved are based outside Europe, and the company publishes the full list on its website.

You do not have to agree. If you would rather not appear, tell me and I will not film while you are here, or I will record somewhere else. You can also change your mind later, and I will have the video removed.

If you want to know how your data is handled or ask for it to be deleted, the address is privacy@egovista.app.

If the person is going to be a subject of the video rather than someone passing through, this notice is not enough on its own. You also need the signed image release in Annex 1 of the Terms.

Annex B — Before you press record

A short check, in the order that saves the most wasted takes.

Read the mission page and note the environment, the length, the minimum resolution and frame rate, the mission code, and the opening movement.

Set your camera to at least the resolution and frame rate the mission states, and wipe the lens.

Look at who is around, and listen. Tell anyone who may appear or be heard what the recording is for, using Annex A, and do not record if someone objects. Make sure no one under 18 can enter the frame.

Look at what is around. Screens, documents, letters, medication, anything that reveals something about someone. Anything belonging to a customer or covered by confidentiality. Move it or turn it away.

Check the light on the working area and record ten seconds as a test. Watch them back. If you cannot see clearly what your hands are doing, neither can we.

Show the mission code to the camera, perform the opening movement, then begin the task in one continuous take.